26 JOURNAL OF GEOLOGY— SUP PLEMBNT 



on the liquid is similar to that produced by zoning, for zoning 

 accomplishes a cutting off of an early crystal from the liquid by the 

 formation of a zone of different composition about it. 



A mass of liquid of composition M is cooled uniformly in all 

 parts and forsterite crystals are imagined to sink from the upper 

 to the lower layers. When the temperature K is reached, the 

 liquid has the composition K in all parts, but it holds only a few 

 olivine (forsterite) crystals in its upper layers and many in its 

 lower layers.^ The total composition of a certain upper layer 

 could, then, be represented by the point D, and of a certain lower 

 layer by some point between M and A. Imagine, for simplicity, 

 that further sinking of crystals is prevented; then, as the tempera- 

 ture falls below K the separation of pyroxene and the re-solution 

 of forsterite proceed in all layers. When the temperature F is 

 reached, the solution of forsterite is complete in the upper layer of 

 composition D, but is still incomplete in the lower layer. With 

 further cooling, re-solution of forsterite continues in the lower layer 

 until at the temperature N all the liquid is used up. The lower layer 

 is now completely crystalline, consisting of a mixture of pyroxene P 

 and forsterite, hut the upper layer is still partly liquid, for, in the 

 meantime, the composition of liquid in this layer has crossed the 

 pyroxene field on the curve FE and the liquid is not entirely used 

 up until the temperature G, much lower than N, is reached. When 

 the temperature G is reached, the upper layer then consists of 

 pyroxene of composition T and tridymite. There is, therefore, 

 not merely a marked difference in composition between the lower 

 and the upper parts of the crystalline product, but also a partly 

 diferent assemblage of minerals. 



It is important to note that as the temperature falls below 

 that of the point F the composition of the liquid portion, which up 

 to this point has been uniform from top to bottom, now pursues a 

 different course in the upper layers from that followed in the lower 

 layers. At a temperature a few degrees below F the composition of 

 the liquid in the lower layers is represented by a point between F 



^ This sinking of olivine crystals is not an imaginary process but has been obtained 

 in crucibles containing these melts (N. L. Bowen, Am. Jour. Sci. [4], XXXIX [1915], 

 17s). 



